Does it PAY to Reimburse Providers Based on Patient Outcomes?

The introduction of Accountable Care Organizations (ACO) to behavioral health – including treatment for individuals with substance use disorders (SUD) and mental illness – is likely to incentivize providing high quality care in a cost efficient manner.

A related health care policy involves the use of “pay-for-performance” (P4P) systems, where providers are paid for services depending on outcomes of patients rather than services rendered. The idea behind this initiative is that incentivizing clinicians to produce good outcomes will lead to better patient care and enhanced remission and recovery rates.

Research on how such approaches impact treatment and recovery outcomes could have significant implications for those systems as well as for providers within Accountable Care Organizations.

Understanding empirically how the implementation of these systems influences treatment and recovery can inform modifications at the micro-level (e.g., patient care) and macro-levels (e.g., treatment program approaches and local and state behavioral health policies).

In the current study, Mason et al. evaluated the impact of a pilot P4P program within 8 SUD treatment centers in the United Kingdom during 2012-2013.


Primary outcomes were:

  1. “successful treatment completion”, defined as discharge from treatment abstinent from one’s substance of choice as well as heroin and crack-cocaine
  2. “declining to continue with treatment”, often referred to as treatment drop-out or attrition.


It is worth mentioning that treatment centers were paid not only for “successful treatment completion” but also if patients were not re-admitted to treatment or to the criminal justice system within 12 months after discharge. However, these latter two payment criteria were not examined in the current study.

To measure the impact of the pilot program on successful treatment completion and treatment drop-out, researchers used a “difference-in-difference” analysis, measuring the difference in outcomes before and after the pilot program was implemented among the 8 participating sites and subtracting the difference during the same time period among the 141 non-participating sites.


They also conducted several supplemental analyses comparing outcomes of patients from the eight participating programs to:



As non-participating programs could also adopt the P4P system, authors also examined differences between P4P and non-P4P sites. They performed analyses at the level of the individual (N = 167,372 in non-participating sites and N = 11,488 in participating sites during 2012-2013), controlling not only for the similarity in patients that was accounted for by attending the same program (“proportion of total variance accounted for by within-site correlation”), but also for several other individual factors, including demographic characteristics, substance use profiles, and having been referred by the criminal justice system.

Results showed that sites in the P4P program did worse than non-participating sites. Specifically, non-participating sites had 1.3% fewer patients successfully completing treatment, and 0.9% more patients drop out of treatment. These findings were consistent across each of the supplementary analyses.


This policy analysis of a pay-for-performance (P4P) pilot program showed that paying treatment programs based on patient outcomes may have a negative impact on treatment and recovery outcomes.


This was unexpected as the point of the program was to enhance patients’ outcomes.

Although the actual difference was somewhat small in magnitude, a reliable difference of this size in tens of thousands of patients could have a notable negative effect from a broader public health perspective. Authors speculated that, among P4P programs, receiving less money for a patient if he/she is readmitted to treatment within the year after discharge may have explained the lower proportion of “successful treatment completions” among patients attending the pilot program sites. Specifically, a provider from a P4P program may have been less likely to discharge a patient, even if doing well, if there were any concerns that the patient might relapse and require treatment re-admission within a year (e.g., due to multiple drug users in their social network).

They also hypothesized that the nature of treatment may have changed in relation to the P4P system leading to greater likelihood of treatment drop-out. Although they do not speculate further, one example may be that patients in a publicly-funded health care system are turned off by providers that emphasize results over process.


  • For individuals & families seeking recovery: Although more research is needed, it may be important to understand how treatment programs are reimbursed for their services, and whether the types of treatments they provide fit well with your needs.
  • For scientists: Follow-ups to this important study of health care policy might consider qualitative investigation to help better understand the motivations and experiences of program managers and clinicians in this type of initiative. Further research examining broader outcomes (e.g., treatment readmission) with longer follow-ups is also needed.
  • For policy makers: More research is needed to understand the influence of P4P policies on providers and patient outcomes before definitive policy implications can be formulated.
  • For treatment professionals and treatment systems: Preliminary data suggests more work is needed to understand how and whether P4P programs are feasible, cost-effective, and improve patients’ outcomes.


Mason, T., Sutton, M., Whittaker, W., McSweeney, T., Millar, T., Donmall, M., . . . Pierce, M. (2015). The impact of paying treatment providers for outcomes: difference-in-differences analysis of the ‘payment by results for drugs recovery’ pilot. Addiction. doi: 10.1111/add.12920